Web Analytics and KPIs: From Measurement Plan to Decision

Web analytics helps you when you decide in advance what your website is supposed to achieve: goals become KPIs, KPIs become a documented measurement plan that you implement technically with Google Analytics 4, Search Console and CRM – in compliance with data protection and with visitors’ consent. That way clicks become traceable connections all the way to the order.

Most companies have more data than they use and still cannot say which channel brings inquiries. This article shows how to build a measurement plan, think consent and data quality together, attribute channels and build dashboards that make decisions easier.

Goals and measurement plan: the foundation of any web analytics

Without defined goals, analytics stays descriptive instead of steering. Before any tracking script is installed, define what your website is supposed to achieve.

Macro and micro conversions

• Macro conversions are the actual business goals: an inquiry, an order, an appointment booking.
• Micro conversions are intermediate steps that indicate interest: a download, watching an explainer video, a click on the phone number, opening the contact form.

Micro conversions are especially valuable when your offering needs explanation and decisions take longer: they show early whether content and user guidance work, long before enough inquiries exist for reliable statements. A machine builder, for example, receives only a few inquiries per month. If the number of downloads of technical data sheets rises, though, that is an early signal that a campaign reaches the right audience.

Assigning KPIs to roles

Role	Guiding question	Typical KPIs
Management	How many inquiries and orders come in, at what cost?	Inquiries, cost per inquiry by channel, revenue from web inquiries
Marketing	Which campaigns and content work?	Sessions and conversions per source, conversion rate of key pages, newsletter sign-ups
Sales	Which inquiries are qualified?	Qualified inquiries, closing rate per source (from the CRM)
Web team	Where do users drop off?	Form drop-offs, load times, error rate

Avoid vanity metrics with no business connection. Rising page views are pleasing but say little if the number of inquiries stays the same. A good check question for every metric is: Which decision would we make differently if this value changed?

Documenting the measurement plan

Record which events are measured, what they are called, which parameters they contain and which goal they serve. This measurement plan is the shared language of marketing, development and management and prevents nobody from knowing what is actually being measured after every website update. How to set up events in Google Analytics 4 in practice is described in the article Configuring GA4 correctly.

Tools: GA4, Search Console and CRM working together

Most company websites get by with a few tools, which you should connect with one another:

Tool	Purpose	Note
Google Analytics 4	Visits, events, key events	Detail data for explorations is kept for only 2 or 14 months – export important analyses
Google Tag Manager	Manage tags and consent control centrally	Version changes and test before going live
Search Console	Search queries, clicks, indexing	Data is available for only 16 months; link with GA4
CRM	Qualification and orders	Store the source of every inquiry, attribute closings
Ad accounts	Costs and clicks	Evaluate costs and results together

In GA4, conversions have been called key events since 2024. There, the bounce rate is the share of sessions without interaction; a session counts as engaged if it lasts at least ten seconds, triggers a key event or includes at least two page views. Figures from old Universal Analytics reports can therefore not be compared one to one. The glossary offers an introduction under Google Analytics 4.

GDPR, consent and data quality

Tracking needs a sound legal basis. If a tool stores or reads information on the device, consent is generally required under Section 25 of the TDDDG (the German Telecommunications Digital Services Data Protection Act); the GDPR rules additionally apply to the subsequent processing of personal data. The consent banner must be connected correctly from a technical standpoint: scripts may only load after consent has been given, and a refusal must actually be respected.

• Consent Mode v2: passes the consent status to Google services; since 2024 Google requires it for users in the European Economic Area when measurement and advertising features are used. In basic mode, tags load only after consent; in advanced mode they first send cookieless signals that Google uses for modeling. Whether that is acceptable for you under data protection law is something to clarify with your data protection advisor.
• Documentation: record data flows, maintain data processing agreements, put data transfers to third countries on a sound legal footing.
• Alternatives: data-minimizing tools such as a self-hosted Matomo can simplify matters. Whether consent is unnecessary in a specific case depends on the configuration and the legal assessment and must be checked.
• Server-side tagging: can improve control over data flows, but does not replace consent.

Data gaps are a reality: visitors decline, browsers and ad blockers suppress measurements. Interpret trends rather than absolute figures. You will find more detail in the article GDPR, tracking and trust and in the glossary under Google Consent Mode v2.

Checking data quality regularly

Even cleanly set-up tracking becomes outdated. New pages, changed forms or a change of consent tool interrupt measurements unnoticed. After every major update, check whether the most important events still arrive, and exclude internal traffic and test environments from reporting.

Attribution: which channel worked?

A prospect sees an ad, later reads a blog article, subscribes to the newsletter and finally inquires via the homepage. Which channel gets the credit? Attribution models try to answer this; in GA4 the main choices are data-driven attribution and last-click attribution. No model is perfect. More important than the “right” model is consistency: choose one, stick with it and document the assumptions. In practice, four things help:

• UTM parameters on all links in newsletters, ads and posts, with consistent spelling – the tools distinguish between upper and lower case
• Self-reporting: a question in the contact form – “How did you hear about us?” – captures referrals, trade-fair conversations and hints that no tracking sees
• CRM link: store each inquiry’s source in the CRM and report closings back; for Google Ads this works via offline conversion import, likewise only on the basis of consent
• AI assistants as a source of their own: visits from AI assistants often appear as a referral from the respective domain but are not captured completely – watch them as a trend

Dashboards that make decisions easier

A good dashboard answers the most important questions at a glance. A bad one shows all available data and leaves the viewer on their own.

• Few metrics per dashboard: focus on what influences a decision.
• Show comparison values: previous period or previous year make developments visible.
• Provide context: note campaign launches, website updates or seasonal effects.
• Connect data sources: combine analytics, CRM and ad accounts to see costs and results together.

To get started, free tools such as Looker Studio connected to GA4 and Search Console are often enough. If standard tools reach their limits, for example because data from ERP, CRM, ad accounts and web analytics is meant to flow together, a custom dashboard is worthwhile. One example from our project practice is an AI-supported analytics dashboard for mid-sized companies that brings together metrics from different sources in one place.

Evaluation as a fixed routine

The best dashboard is useless if nobody looks at it. Establish a fixed meeting – for example monthly – where marketing, sales and management discuss the most important metrics. Derive concrete actions from each round and check their effect at the next meeting.

Practical example: cost per order instead of clicks

A worked example shows why the link to the CRM matters. A service provider invests €1,800 in Google Ads in one month and receives 12 inquiries via the form. Web analytics reports €150 per inquiry. The CRM shows: those 12 inquiries produced 3 orders – that is €600 in advertising cost per order. In the same month the newsletter generates 4 inquiries and 2 orders, at lower media cost but with effort for editing and sending. Only the combination of both data sources shows which channel works better economically.

The figures are a worked example for illustration, not measured values. Without UTM parameters, newsletter traffic could easily have been counted as “direct” in the reporting – and the comparison would not have been possible.

Typical mistakes and checklist

Typical mistakes in web analytics

• Measuring everything, evaluating nothing: countless events, but no clear questions for the data
• Inconsistent naming: the same actions are called different things in different parts of the website
• Missing campaign parameters: links in newsletters and ads without UTM parameters end up in the wrong channel in reporting
• No connection to the CRM: inquiries are counted, but nobody knows which of them became orders
• A banner without technical effect: scripts load despite refusal – a data protection and trust risk

Checklist: web analytics in seven steps

1. Define the website’s business goals
2. Define macro and micro conversions
3. Assign KPIs to roles in the company
4. Document the measurement plan with events and naming conventions
5. Connect and test the consent banner and tracking cleanly from a technical standpoint
6. Choose an attribution model, set UTM rules, record assumptions and link the CRM
7. Build dashboards per audience and discuss them together regularly

How Nano Concepts supports you

Nano Concepts implements tracking, interfaces and custom dashboards technically – as part of our web development and dashboard development, which brings metrics from ERP, CRM, store, web analytics and Excel together in one overview. We develop campaign evaluation and marketing strategy together with our partner agency LOMA DESIGN so that numbers turn into concrete actions. A lean dashboard with few sources is possible in the Starter package from €2,500; in the free initial consultation we clarify which questions your numbers should answer.

This article does not replace legal advice.